January 27, 2020
January 27, 2020
by Anastasiya Parkhomenko | 5 min read
All those who work in the challenging maritime industry surely know how important it is to plan and carry out regular equipment maintenance. Not only may a vessel experience unexpected breakage while cruising unfriendly waters, but also such unplanned downtime can result in million-dollar losses due to undelivered goods or services and a company’s damaged reputation.
Traditional maintenance approaches include:
Conventional maintenance strategies suppose that machines are overhauled or replaced according to empirical values reflecting actual usage of a part or at predefined intervals of time. Often the consequence of such approach is unnecessary replacement of components that may still be fully functional. This results in excessive docking times and limited availability of a vessel. Moreover, worn-out assets may not be overhauled as they are not yet believed to fail according to their estimated lifecycle.
Thus, with predictable availability being the key prerequisite to thrive in a competitive maritime sector, it is imprudent to rely on planned or periodic maintenance as well as on regular inspections commonly run by the vast majority of companies nowadays. Equipment and machinery, especially the sea-based ones, are subject to all kinds of damages and failures that are usually also very expensive to repair. That’s why many international companies are experiencing an important paradigm shift: from passivity to proactivity, from planned maintenance to… predictive maintenance.
Predictive maintenance is undoubtedly the best way to cope with on-the-spot failures by using an innovative early detection system which sends an alert while machinery is still functioning normally. It secures lower risk of cascaded damages and enhanced reachability as the cost of repairing a ship in the port between voyages is significantly lower than one inflicted by a sudden failure on a remote route, ceteris paribus. Moreover, it is crucial for stakeholders to be aware of the equipment condition both before leaving the port as well as during the upcoming voyage in order to provide a well-orchestrated maintenance and guarantee travel safety. ‘Predictivity’ means enhancing the condition-based maintenance by using nonintrusive analysis of equipment or machinery, which relies on wealth of real-time data collected from monitoring systems and sensors in order to provide a reliable forecast when components need to be maintained. Such forecast is based on the numerous parameters such as temperature, engine usage, wear and tear, weather and many other factors in order to precisely predict overhauls or replacements necessity while ensuring continuous runtime of highly integrated maritime systems.
During the forecast period of 2017 to 2024, the global predictive maintenance market is estimated to grow fast with a CAGR of 29.1 percent from 1.6 billion US dollars in 2016. According to the ‘Data Bridge Market Research’, the major market drivers are:
Remarkably, predictive maintenance is estimated to increase productivity by 25%, reduce breakdowns by 70% and lower maintenance costs by 25%, according to Deloitte Analytics Institute.
There are some advanced technologies underlying this revolution.
Internet of Things allows sensors and devices to be part of a network, measuring and sending their respective parameters to analytical systems and Artificial Intelligence platforms. If data sent by connected objects can only lay down the status quo, it will be important then to process data by means of analytics in order to detect eventual anomalies.
Studying how different mechanisms may lead to anomalies empowers us to spot their causes before adverse effects appear. This is possible due to predictive models, specifically classification algorithms and anomaly detection, which are two well-established techniques of Machine Learning. Classification algorithms are very powerful tools to differentiate normal operations from the erroneous ones based on abundant historic data providing sufficient examples of previously recorded failures. Anomaly detection methods specialise in identifying outliers among a plethora of normal operations. Then algorithms used for predictive maintenance are parallelised and streaming architectures for Big Data ensure extremely fast real-time processing of the incoming data in the below seconds regime.
As the result of a joint effort with one of the world’s largest shipyards with clients in the cruise, the defence and the cargo industries, and Fincantieri S.p.A., an integrated digital platform combining Big Data and IoT technologies will be soon built up by KDM Force Ltd. A standardised connected vessel platform will enable stakeholders to collect, secure, buffer, transport and pre-process operational data coming from a vessel to the cloud or an on-premises data centre.
In order to ensure effective fleet management, the system uses multi-dimensional data supplied from numerous sensors on board that consents early detection of any anomalies on-board before they have a grave impact on a vessel’s performance. During the selection process of the parts to be monitored, it is essential to differentiate between critical and non-critical system components; this is particularly valid for maritime technologically sophisticated systems with a plethora of mechanisms. Accurate assessment, classification and failure rate estimates of the critical components assure the predictive power of the forecast. Once the installed sensors generate data automatically, the biggest challenge is to close the networking loop between a vessel, shore, a supplier, a yard and the client.
The ‘edge computing’ architecture is used to improve response times and save bandwidth as the connection in the open sea can be extremely limited. Consequently, the platform will run in a cloud environment while the hardware on board the connected vessels will also be given some computation and storage power so the full data will be recorded and uploaded once a vessel is docked. Thus, clients will receive and manage data in an easy and well-orchestrated way that allows them to take effective decisions in a timely manner. Maintenance can be performed in a just-in-time manner and associated logistic processes can be scheduled in advance. Moreover, such process will accelerate the clients’ digital transformation by dramatically enhancing their capabilities in the digital domain.
Imagine a software solution which is able to incorporate data from different sources (e.g. AIS, weather and wave height data) and that can exchange anonymised data via a stable and well documented API. Moreover, other proprietary applications can be built on top of this software solution. These applications can be developed by the yards, ship owners, suppliers or independent software companies. Thanks to vendor-neutral, open source EdgeX Foundry architecture, interoperability will be one of the top advantages of our solution, as well as the possibility to collect data by using OT and IT protocols even if there is no internet signal available. Not only will a connected vessel let our clients know as soon as there are any alarming events in its operational status, it may even request a docking space, define a list of necessary spare parts and provide the shipyard with all information needed for a fast and efficient maintenance.
New and innovative business use cases around predictive maintenance will surely emerge as unexpected maintenance leads to extensive downtimes and jeopardises the production chain across the ever-growing and increasingly interconnected maritime industry. KDM Force embraces the challenge to help smart vessels maximise their own availability while further reducing the risk of breakdowns leveraging the power of Big Data and cutting-edge technology.